PROPERTIES OF M31. IV. CANDIDATE LUMINOUS BLUE VARIABLES FROM PANDROMEDA

Luminous blue variables (LBVs) are hot massive stars that undergo sporadic eruptions on timescales of years and decades (Humphreys & Davidson 1994). The prototype is S Doradus, as well as Hubble–Sandage variables in M31 and M33 (Hubble & Sandage 1953), which shows eruptions of 1–2 mag level over several decades. Other examples are η Carina and P Cygni, which show giant eruptions (>2 mag) over several centuries. Conti (1984) was the first to coin the name "luminous blue variables" for this type of star, and separated them from other type of bright blue stars, such as Wolf–Rayet stars.

LBVs play an important role in a very late stage of massive star evolution. They are considered as a transition phase where O stars evolve toward Wolf–Rayet stars (Meynet et al. 2011). LBVs were originally regarded as supernova (SN) impostors because they often show giant eruptions mimicking the explosion of SNe, but the central star remains after the ejecta have been expelled. However, a link between LBVs and SN progenitors was suggested by Kotak & Vink (2006) when interpreting the radio light curves of SNe. The radio emission seen after the SN explosion is induced by the interaction between SN ejecta and the progenitor's circumstellar medium, thus radio light curves bear information on the mass-loss history of the progenitor. Kotak & Vink (2006) suggested that radio light curves of SNe indicate high mass-loss histories of the progenitors, which matches well with LBVs.

Pre-eruption images of several SNe also suggest LBVs as their progenitor. For example, the progenitor of SN 1987A was recognized as a blue supergiant (Walborn et al. 1987) and Smith (2007) suggested that it could be classified as a low-luminosity LBV. Gal-Yam & Leonard (2009) identified the progenitor of SN 2005gl using Hubble Space Telescope (HST) and considered it to be a LBV. Recently, a previously known LBV, SN 2009ip, underwent its third eruption and has been linked to a true SN (Mauerhan et al. 2013). The nature of the recent eruption of SN 2009ip is under debate; subsequent follow-up has been carried out to verify or reject it as a core-collapse SN (Fraser et al. 2013). However, there are only a few known LBVs, either in our Galaxy or in M31 and M33, thus, increasing the number of known LBVs is essential for understanding their nature and evolution.

In addition to the pioneering decades-long photometric monitoring campaign conducted by Hubble & Sandage (1953), there are several methods to uncover LBVs. For example, LBVs are strong UV and Hα emitters (see Massey et al. 2007, and reference therein) and can be revealed, e.g., with observations of the Galaxy Evolution Explorer (GALEX) satellite or Hα surveys. Massey et al. (2007) conducted a Hα survey of M31 and M33 and spectroscopically followed up on a selected sample of strong Hα emitters. By comparing the spectra of their Hα emitter sample with known LBVs, they were able to identify candidate LBVs, which saved a substantial amount of time required to photometrically uncover LBVs. Because they have uncovered more than 2500 Hα-emitting stellar objects, they can only follow up on dozens of them, yet there are many more to be explored. Humphreys et al. (2013) are currently exploring other Hα-emitting sources on this list, in combination with infrared photometry including Two Micron All Sky Survey (2MASS), Spitzer, and WISE to search for luminous and variable stars. Since LBVs undergo several eruptions and exhibit high mass-loss rates, they accumulate vast amounts of material in their circumstellar environment which could be detectable in the infrared (e.g., Gvaramadze et al. 2012). Khan et al. (2013) have made use of Spitzer/IRAC photometry and searched for η Carina analogs in nearby galaxies including M33 (but not M31). They estimate that 6 ± 6 of their candidates are true η Carina-like sources.

Here we outlined a novel approach utilizing mid-term photometric variation, as well as optical and infrared color to search for LBVs using the Local Group Galaxy Survey (LGGS) optical and 2MASS infrared photometry, with the combination of the photometric variability from the PAndromeda monitoring campaign. Our paper is organized as follows. In Section 2, we describe the optical and infrared data we use. In Section 3, we outline our method. A discussion of our candidates is presented in Section 4, followed by an outlook in Section 5.

2.1. Optical Data

The time-series photometric data employed to search for variability are from the PAndromeda project. PAndromeda monitored the Andromeda galaxy with the 1.8 m PS1 telescope with a ∼7 deg² field-of-view (see Kaiser et al. 2010; Hodapp et al. 2004; Tonry & Onaka 2009, for a detailed description of the PS1 system, optical design, and the imager). Observations have been taken in the r_P1 and i_P1 filters on daily basis during July to December in 2010, 2011, and 2013 in order to search for microlensing events and variables. The distribution of the observations in the r_P1-filter is shown in Figure 1. Several exposures in g_P1, z_P1, and y_P1 are also taken as complementary information for studies of the stellar content of M31.

Zoom In Zoom Out Reset image size

The data reduction is based on the MDia tool (Koppenhoefer et al. 2013) and is explained in Lee et al. (2012) in detail.

We outline our data reduction steps as follows. The raw data are detrended by the image processing pipeline (Magnier 2006) and warped to a sky-based image plane (so-called skycells).

The images at the skycell stage are further analyzed by our sophisticated imaging subtraction pipeline mupipe (Gössl & Riffeser 2002) based on the idea of image differencing analysis advocated by Alard & Lupton (1998). This includes the creation of deep reference images from the best seeing data, stacking of observations within one visit to have a better signal-to-noise ratio (hereafter "visit stacks"), subtraction of visit stacks from the reference images to search for variabilities, and creating light curves from the subtracted images.

We have shown in Kodric et al. (2013) how to obtain light curves for resolved sources from the PAndromeda data. The major difference of the data used in this work is that our present data set contains 3 yr of PAndromeda data instead of 1 yr and a few days from the second year of data used in Kodric et al. (2013). The sky tessellation is also different, because the central region of M31 is in the center of a skycell (skycell 045), instead of at the corner of adjacent skycells (skycell numbers 065, 066, 077, and 078) as in Kodric et al. (2013); the skycells are larger and overlap in the new tessellation. The new tessellation is drawn in Figure 1 of Lee et al. (2013). We have extended the analysis to 47 skycells, twice as many as the number of skycells used in Kodric et al. (2013). The skycells we used are 012–017, 022–028, 032–038, 042–048, 052–058, 062–068, and 072–077, which cover a 7 deg² area of M31. The search of variability is conducted in both r_P1 and i_P1, where we start from the resolved sources in the r_P1 reference images and check for variability in both r_P1 and i_P1 filters.

In addition, we also use the deep photometric catalog from the LGGS (Massey et al. 2006). The LGGS utilized the 4 m KPNO telescope to observe the M31 galaxy. Their Mosaic CCD camera has a resolution of 0261 pixel⁻¹ at the center which decreases to 0245 pixel⁻¹ toward the corner. The field-of-view of the camera is roughly 36' × 36'. M31 was observed between 2000 and 2002 in Johnson UBVRI filters with seeing values from 08 to 14. The observations cover 10 fields, corresponding to 2.2 deg² along the major axis of M31 (see Figure 1 of Massey et al. 2006).

The astrometric solution for each frame was derived by matching with the USNO-B1.0 catalog (Monet et al. 2003). The point-spread function (PSF) photometry in each filter was obtained with the IRAF DAOPHOT routine, and calibrated against the Lowell 1.1 m data with zero points and color terms. The final LGGS catalog contains 371,781 stars in M31, with 1%–2% statistic error at 21 mag and 10% at 23 mag. When matching the LGGS catalog to the PAndromeda catalog, we found a median astrometric difference of 036.

2.2. Infrared Data

We designed our selection algorithm based on the optical and infrared properties of known LBVs.

The first criterion is the optical color. As can be seen in Figure 2, the known LBVs and LBV candidates presented by Massey et al. (2007) are rather blue in the B − V versus V color–magnitude diagram (CMD). We thus set the criterion that B − V < 0.5 mag and V brighter than 20 mag to select for optically blue and luminous objects. This enables us to filter out possible contaminations from foreground stars and unresolved background galaxies.

Zoom In Zoom Out Reset image size

The next criterion utilizes the optical variability from PAndromeda data. Since OB stars also show aperiodic variation at the <0.1 mag level, the so-called α Cyg variables (van Leeuwen et al. 1998), and since Clark et al. (2012) have accounted for a variation of ⩽0.4 in magnitude in their M33 LBV candidates by similar mechanism, we require light curve variations of >0.4 mag from r_P1 to secure mid-term LBV variability. Relaxing this criterion might allow us to find more candidates. We will come to this point in Section 4.4. Occasionally there are a few data points that have large error bars that deviate from other observations. Such outlier measurements could render a light curve with Δ mag > 0.4 and leads to false detection. To reduce the number of false detections, we thus require that at least 25 data points vary at the 10σ level with respect to the mean value from all the measurements in a single light curve.

The last criterion is set by the infrared color. Kraus et al. (2014) have indicated that B[e] supergiants (B[e]SGs) also show variabilities similar to LBVs, and pointed out that LBVs and B[e]SGs can be distinguished from the infrared colors as shown in a recent study by Oksala et al. (2013) using samples from the Milky Way and Magellanic Clouds. We plot the sample of Oksala et al. (2013) in Figure 3, as well as the four known LBVs in M31. The M31 LBVs follow the distribution of Galactic LBVs; most of them have H − K < 0.5. We thus set an upper H − K limit of 0.5 to select candidate LBVs.

Zoom In Zoom Out Reset image size

There are seven sources that pass the aforementioned criteria. We inspect their light curves individually and select four as candidate LBVs. The remaining three sources are all known LBVs (AE And, M31 Var 15, and M31 Var A-1). The only known LBV that does not pass our selection criteria is AF And, due to its variability smaller than the 0.4 mag criterion (0.375) during the time span of PS1.

The number of stars that passed each criterion is listed in Table 1. There are in total four previously unknown sources that pass all of our criteria. Their optical and infrared photometry, as well as their spatial distribution and light curves, are shown in Table 2 and Figures 4 and 5, respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We discuss these sources in the following section.

In this section, we examine the properties of our LBV candidates, investigate whether they are UV emitters, derive their ages from the massive star evolutionary model, and examine their HST images (if available).

4.1. Galaxy Evolution Explorer UV Detection

It has been suggested that LBVs can be revealed in the UV channel (Massey et al. 1996). We have plotted the positions of our candidates on the GALEX near-UV images (GALEX nearby galaxy atlas; Gil de Paz et al. 2007). As shown in Figure 6, all LBV candidates are aligned with bright UV sources, and they are located in the spiral arms of M31 (see Figure 4). For comparison, we also show close-up views of the known LBV and LBV candidates listed in Massey et al. (2007) in the Appendix.

Zoom In Zoom Out Reset image size

4.2. Comparison with Isochrones

In order to see whether our candidates are consistent with the evolutionary model of massive stars, we compare the B − V color and the V band magnitude of these four candidates with the latest Geneva evolutionary tracks (Ekström et al. 2012). As shown in Figure 7, our candidates are in good agreement with the Geneva models. In addition, from the evolutionary tracks, we are able to estimate their ages. As indicated by the models, their ages are of the order of 10⁷ yr.

Zoom In Zoom Out Reset image size

In Figure 7, we also indicate the possible variabilities of LBVs by drawing the photometric variations $\Delta\, {\rm mag}_{r_{\rm P1}}$ seen from PAndromeda as an error bar. LBVs can also suffer from dust extinction from their circumstellar material. To take this into account, we apply a correction for the extinction effect on B − V color using the extinction map by Montalto et al. (2009). By assuming A_V/E(B − V) = 3.1, we also correct the extinction effect on the V-band magnitude. Taking the extinction effect into consideration, the true B − V value of our LBV candidates would be smaller. In this case, our LBV candidates would be blueward on the color–magnitude plot in Figure 7, which is still consistent with the evolutionary model of an age on the order of 10 Myr.

4.3. Hubble Space Telescope Observations

To confirm that our LBV candidates are stellar objects, we thus request M31 HST images from the Panchromatic Hubble Andromeda Treasury project (PHAT; Dalcanton et al. 2012). Since the PHAT survey only covers the northern disk of M31, we only find images for two of our LBV candidates, PSO J11.0457+41.5548 and PSOJ11.2574+42.0498. We show the HST Advanced Camera for Surveys images of them in Figure 8. With the exquisite spatial resolution of HST, we can see that the shape of PSO J11.0457+41.5548 and PSOJ11.2574+42.0498 traces the typical PSF in the field.

Zoom In Zoom Out Reset image size

In addition to the PHAT archived images, we also found PSO J10.1165+40.7082 covered by the "Treasury Imaging of Star Forming Regions in the Local Group" program (Bianchi et al. 2012). The HST images of this candidate, astrometrically aligned to the PAndromeda data using our own pipeline (M. Kodric et al. in preparation), is shown in Figure 8 as well. HST resolved two sources within 1 arcsec of PSO J10.1165+40.7082. To distinguish the varying source, we examine the difference image from PAndromeda during maximum flux (at MJD = 55816.38 and 56218.38) and found that the brighter source in the F814W band is the varying source.

4.4. A Further Look at the Variability Criterion

In Figure 9, we plot the number of sources that pass our optical and infrared photometric criteria against the photometric variability from PS1 r_P1-band light curves. In total, there are seven sources showing $\Delta\, {\rm mag}_{r_{\rm P1}} > 0.4$ mag, four of which we selected as possible LBV candidates. The remaining three are all known LBVs (AF And, M31 Var 15, and M31 Var A-1). In Figure 9, there are three additional sources that vary at the 0.3 mag level, which are AE And with $\Delta\, {\rm mag}_{r_{\rm P1}}$ = 0.375 and other variables with $\Delta\, {\rm mag}_{r_{\rm P1}}$ = 0.339 and 0.307. Even if we lowered the $\Delta\, {\rm mag}_{r_{\rm P1}}$ criterion to the lowest value of known LBVs (0.375), we would only select AE And, but no additional new LBV candidates.

Zoom In Zoom Out Reset image size

For comparison, we also show the r_P1 light curves of the four known LBVs listed in Massey et al. (2007) in the Appendix.

We study the photometry of known M31 LBVs from Massey et al. (2007) and present a new approach to search for LBVs using optical and infrared information. We have selected four candidate LBVs sharing the same properties of known LBVs in terms of optical and infrared colors; they are also observed in the GALEX UV data and all of them are located within M31 spiral arms. Our sample exhibits photometric variation >0.4 mag as seen from the PAndromeda survey. These sources are in agreement with the stellar evolution model of the Geneva group, which gives an age between 10 and 100 Myr. This implies that while low mass stars are still in the early stage of evolution in the spiral arms, massive stars have already evolved into the LBV stage. Though the true nature of our sample awaits a spectroscopic confirmation, the bright UV emission, optical and infrared colors, photometric variability, and the HST image altogether indicate that our candidates are very likely LBVs.

We will request spectroscopic observation to confirm that our candidates are true LBVs, and classify them according to the taxonomy scheme outlined by Massey et al. (2007). In addition, it has been shown that LBVs can be surrounded by nebula with dust (Vamvatira-Nakou et al. 2013). Spectra in the mid-infrared will help us to probe dusts with colder temperatures; Waters et al. (1997) have used the short wavelength spectrometer on board Infrared Space Observatory to obtain mid-infrared spectra of several known LBVs and led to the discovery of cold circumstellar dust with a temperature about ∼50 K. Future space observatories such as the SPICA telescope (Nakagawa & SPICA Working Group 2001) will also provide spectra in the wavelength of mid-infrared.

We thank the referee for useful comments. This work was supported by the DFG cluster of excellence Origin and Structure of the Universe (www.universe-cluster.de).

The Pan-STARRS1 Surveys (PS1) have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under grant No. AST-1238877, and the University of Maryland.

This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation.

This research has made use of the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

GALEX (Galaxy Evolution Explorer) is a NASA Small Explorer, launched in 2003 April. We gratefully acknowledge NASA's support for construction, operation, and science analysis for the GALEX mission, developed in cooperation with the Centre National d'Etudes Spatiales of France and the Korean Ministry of Science and Technology.

This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France.

In this section, we present information of LBV and LBV candidates from Massey et al. (2006), as supporting material to show that our LBV candidates have similar properties to known LBVs.

In Table 3, we collect the V-band magnitude, the B − V color, and the photometric variability in the r_P1 light curve of four known LBVs in the Massey et al. (2007) sample.

A.1. Galaxy Evolution Explorer UV Detection

We collect UV images of known LBV and LBV candidates listed in Massey et al. (2007) and present in Figures 10 and 11. The images are taken at near- and far-UV (NUV and FUV) by GALEX. Most of the LBV and LBV candidates in Massey et al. (2007) are bright sources in the NUV, but some of them show faint or no FUV emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A.2. Hubble Space Telescope Observations

We search for HST archive images of the known LBVs listed in Massey et al. (2007) in the "Panchromatic Hubble Andromeda Treasury" program (Dalcanton et al. 2012), and found that two of the known LBVs (AF And and M31 Var 15) have been observed by this program. Their images are shown in Figure 12.

Zoom In Zoom Out Reset image size

A.3. Light Curves from PS1 Data

We show the light curves of four known LBVs from Massey et al. (2007) in Figure 13. The light curves are from our PS1 3 yr data. The photometric variation during the 3 yr time span are listed in Table 3.

Zoom In Zoom Out Reset image size